Frequency Domain Measurements: Spectrum Analyzer or Oscilloscope

Keysight Technologies

Frequency Domain Measurements:

Spectrum Analyzer or Oscilloscope?

Introduction

For generations, the rules for RF engineers were simple: frequency-domain measurements (output frequency, band power, signal bandwidth, etc.) were done by a spectrum analyzer, and time domain measurements (pulse width and repetition rate, signal timing, etc.) were done by an oscilloscope.

As the digital revolution made signal processing techniques easier and more widespread, the lines between the two platforms began to blur. Oscilloscopes started incorporating Fast Fourier Transform (FFT) techniques that converted the time-domain traces to the frequency domain. Spectrum analyzers began capturing their data in the time domain and using post-processing to generate displays. Still, there were some clear distinctions between the two platforms. For example, oscilloscopes were limited in sample speed. They could see signals down to DC, but only up to a few GHz. Spectrum analyzers could see high into the microwave range, but they missed transient signals as they swept.

What if you needed to see a signal in the time domain with a carrier frequency of 40 GHz or capture a complete wideband pulse in X-band? As technologies in EW, radar and communications move ahead, the demands on the test equipment become greater.

As more possibilities have opened for RF and microwave equipment because of new digital processing technologies, they have also increased opportunities for test equipment. Spectrum analyzers and oscilloscopes can do much more than they could even a few years ago, and as they expand in capabilities, the lines between them become blurred and sometimes even erased.

Changes in technology

The digital revolution has changed the way spectrum analyzers and oscilloscopes work at a fundamental level:

Spectrum analyzers

Most spectrum analyzers now have an all-digital signal processing (DSP) section. Like a classic spectrum analyzer, the incoming signal is down-converted to a much lower Intermediate Frequency (IF). This signal is sampled, digitized by an Analog-to-Digital Converter (ADC) and processed using DSP techniques. The spectrum analyzer now provides two modes of operation: sweeping the LO to see signals across a wide range or “pausing” the LO to simultaneously see everything within the analysis bandwidth allowed by the sampling rate.

The main advantages of this technique are improved accuracy and reliability: with the analog components replaced by digital processing, the uncertainties inherent in the analog components can be greatly reduced. Classic spectrum analyzer components like Resolution Bandwidth (RBW) filters and log amplifiers are now implemented digitally and made more accurate and repeatable.

Two additional advantages are given by “pausing” the local oscillator and collecting data around one frequency. The first is the ability to view a wideband signal in the time domain. Because the signal is now sampled and digitized, it can be displayed in the time domain, just like an oscilloscope. The main difference here is the spectrum analyzer data has been down-converted, so the displayed data is relative to the center frequency of the measurement. The second is the ability to see the phase information of the signal. By performing some basic DSP, the phase of signals like communications signals and radar chirps can be demodulated and analyzed. Some manufacturers refer to spectrum analyzers with this capability as “signal analyzers” to reflect this new ability to demodulate and analyze the signals riding on the carrier. The signal analyzer still has an important limitation: the sampling speed of the ADC in the digital IF section. This limits the analysis bandwidth to no more than a few hundred MHz.

Oscilloscopes

Like the IF section of a signal analyzer, signals incident on the oscilloscope’s front end are sampled and processed digitally, but at much higher rates. There are now real-time oscilloscopes that can sample at speeds up to 160 GSa/s, allowing the oscilloscope to see signals from DC up to 63 GHz.

There are some considerations that can limit the usefulness of this architecture. One is the avalanche of data that results from such fast sampling. Hundreds of gigabytes of data are generated per second, and typically only fractions of a second can be captured and analyzed at one time if the full bandwidth is used. Data processing techniques like segmented memory can extend the capture time, but this technique only works for pulsed or repeating signals. Another consideration is that high-speed ADCs usually only offer 8 bits of resolution, unlike the 14 or 16 bits in signal analyzers. If you are looking for low-level signals across a wide frequency range (e.g., a spur search), the spectrum analyzer has an advantage, but for most communications and radar applications, the differences between the two boxes may not be significant.